Computerized ECG Machine

Transcription

1 Computerized ECG Machine R.A.D. Aruna Niresh, T.R. Ariyarathne, R. Lelwala and K.D.I Wasudewa* Department Physics, University of Colombo, Colombo 03, Sri Lanka. *Medical Center, University of Colombo, Colombo 03, Sri Lanka. Tel : , Fax: , aruna_niresh@yahoo.com Keywords: Digitizing an ECG signal, Computer based ECG system, Computerized 12 Lead ECG system, Heart infection analyzer. Abstract Main purpose of the research work is to construct a low cost, computer based ECG Machine to be used in ordinary dispensaries Generation of electrocardiogram is done by capturing lead signals through parallel port of the computer. The Computer ECG Machine (CEM) signal detector circuit consists of an instrumentation amplifier, summing amplifiers, Lowpass filter and a lead selector. Microcontroller module has been use to interface the computer and the CEM. The software package developed using MATLAB 5.3 to provide user interface control to the CEM is named as ECGPro. Main task of the software package is to construct standard ECG signal in order to print on a paper and/or display on screen with the grid for visual observation by doctor and to supply digitized signal for computer based DSP analysis, for diagnosis. Implementation of filters and communication with parallel port are also included in the package. Introduction The Computer ECG Machine (CEM) is a system which uses the Standard Clinical ECG technique. Basically, CEM can be separated in to two major parts, namely Hardware Interface (electronic device) and Software Interface (hardware controller and ECG analyzer). parallel port of the PC to send digitized data to the computer. Electrodes Interface circuit Parallel Connection Figure 1.1 Mediums used to interface the human body and the computer Software interface Software Interface itself can be subdivided into two components, namely hardware controller and graphical user interface. Hardware controlling software also consist two parts, control software interface developed using MATLAB 5.3 and the microcontroller program developed using Microchip MPLAB IDE 6.2 software tool. Graphical user interface provides the user with standard feature which comes with the normal ECG machine. Following chapters will briefly describe the features of the software interface Hardware interface Figure 1.1 shows the schematic diagram of interfaces. This provides the interfacing between human body and the computer software (PC). It uses Standard ECG Cables to acquire the voltage variations of the human body and

2 Technical Review Electrocardiogram is a measurement of functionality of human heart. These measurements are taken by placing electrodes on human body and detecting the voltage variations by those electrodes. To generate electrocardiogram, it needs to combine detected signals from the human body and should be digitized in order to send it to the computer. Since this process involves quite a few steps, it is better to discuss them in detail. 2.1 Detection & Generation of ECG It is understood that the amplitude of the detected signal is very small (approximately in rang of 1 mv). Therefore it is necessary to have circuit which has high impedance at the input terminals to obtain the signal without any lost. Also since the 12 ECG Leads (Standard Clinical ECG) is generates using 12 different combinations of signals which are acquired by the 10 electrodes, the circuit should encompass some kind of a selection technique in order to select appropriate electrodes to generate each lead respectively. Since the signal amplitude is very small, noise contribution to the signal is quite significant. The frequency band used in standard clinical ECG technique is 0.05 Hz 100 Hz. Therefore it is reasonable to have a low pass filter before digitizing. But introduction of a low pass filter to the end is not going to eliminate the all noise components. Because there is a major noise contribution from AC power line (50 Hz). Therefore it should also be removed. This can be done by having a 50 Hz notch filter (band reject filter). 2.2 Digitizing the signal Generated ECG pattern is a bipolar signal. Therefore the DC level of the signal should be shifted into a certain level since it is going to use an A/D converter which allows only analogue voltages between 0 V to 5 V. Since digitized data is going to be transferred to the computer, it is important to have a hardware handshaking between sender (interface circuit) and the receiver (computer) for proper transmission of data. 2.3 Control software (ECGPro) This is a combination of both hardware controller software (program) and user interface. Although the hardware controller program is doing the controlling process user interface (front end of the control software) is the main control panel of this device, since this is the interface which going to interact with the end users. Therefore the interface software acts an important roll here. Characteristic features of the interface software can be listed as follows. Patient identification system Signal magnification (as in the standard ECG machine) Changing the mode of recording (manual / automatic) Storing of recorded data Printing of recorded data Back end of the control software is the hardware controller program. Please refer section 4.3 for more details. Design & Implementation The three major sections of CEM have been discussed under the technical review. Therefore it is better to discuss the important steps involved in designing each section separately. 3.1 Designing of signal detector There are 12 different electrocardiograms (12 Leads) to be generated, since this uses standard clinical ECG technique. To generate these 12 leads, it needs to take different combinations of signals which are taken from human body. Therefore it is better to start with the simplest lead, Lead I. 3.2 Designing of ECG amplifier Consider the simplest electrode combination, Lead I. this is a bipolar lead and it measures the voltage difference between left arm and right arm. Following are the techniques that were used to take the difference between two voltage signals. First approach Difference between two signals can be taken by adding one signal to the other by taking phase shift. Figure 3.1 shows the schematic of this approach. But this approach didn t give the desired results. Reason for failing this method is the mismatch of impedances at the input terminals. Because of that reason the signal doesn t match with each other. Also since there is no noise rejection technique, noise is also amplified by the input terminals. Therefore the output of the system fails to give the desired result.

3 Third approach Input 1 Input 2 NonInverting Amplifier Inverting Amplifier Output This approach uses cascaded a low pass active filter (see Figure 3.4) with the previous circuit at the output terminal. Frequency response of the filter is shown under Figure 3.5. Initially the upper cutoff frequency was set to Hz. Figure 3.6 shows the output taken under this approach and it has some improvements over previous outputs. Figure 3.1 Block diagram of the first approach Second approach Actually this approach can be considered as an improved version of the first approach. Instead of using two different amplifier configurations at the input, this method uses two noninverting amplifiers and the outputs drive into a differential amplifier. This circuit forms an instrumentation amplifier. Instrumentation amplifier rejects the common mode signals. Noise can be considered as common mode signals. Therefore noise contribution also can eliminate by using an instrumentation amplifier. Figure 3.2 shows the circuit diagram used under this approach. Figure 3.3 shows the output obtained by the circuit designed under second approach, it has the characteristic variations of a standard ECG wave form for Lead I configuration. Although the required pattern was obtained, still it contains a considerable amount of noise. Maximum frequency interested in electrocardiogram is 100 Hz; therefore it is quite reasonable to have a low pass filter. Figure 3.3 Wave form obtain for Lead I configuration using circuit used in second approach Figure 3.4 Circuit diagram used under third approach. Here the instrumentation amplifier is cascaded with low pass filter. Figure 3.2 Circuit used under second approach Figure 3.5 Frequency response of the low pass filter.

4 10 N R L F V1 V2 V3 V4 V5 V6 R L Lead I Figure 3.6 Output obtained under third approach. This shows how it improved from the previous one. Although there are some noises incorporated in the output, circuit used under third approach was able to produce the desired ECG pattern. Since digital signal processing techniques are going to use, output of the third stage can be accepted. All these circuit setups use TL 084 opamp ICs. Figure 1.6 (chapter 1) shows the different combinations of electrodes which use to generate all 12 electrocardiograms (12 leads). Generation of 12 leads can be done by designing a separate circuit for each lead. But this can be done using some selection technique instead of having separate circuits for each lead. 3.3 Designing of a lead selector First of all it is better to have an over all idea about features which should include this lead selector. Main features of the lead selector can be list as follows, It should be able to take analogue inputs It should be able produce 2 outputs using 12 inputs It should be easy to control Figure 3.7 shows the expanded schematic of all lead combinations. It clearly shows what should be the function of lead selector. By observing the figure 3.7 it is understood that, to generate all 12 leads, an ECG amplifier needs two inputs. By observing figure 3.7, it is understood we can use some kind of a multiplexing method before generating the Lead signals. But it is not possible since the magnitudes of the body signals are very low. Therefore the multiplexing has been done after the generation of the Lead signals. Since each lead signal is going to be digitized, here it is using a 12 to 1 multiplexer to select each lead according to the user requirement. GND R F F L Adder1 R Adder2 L Adder3 F V1 V2 V3 V4 V5 V6 Lead II Lead III avr avl avf Figure 3.7 Detailed block diagram with all lead combinations. sign represents an adder circuit (see section 3.4) and sign represents an ECG amplifier discussed under 3.2. Adder1, Adder2, Adder3 and Addrer4 represent the outputs of corresponding adder circuits. V1 V2 V3 V4 V5 V6

5 3.4 Design of an adder circuit (summing amplifier) Actually here, it uses simple adder circuit which designed using two opamps. Figure 3.8 shows the circuit diagram of the adder circuit. There are four adders used in detector circuit and three of them have 2 inputs only. Therefore number of inputs of the adder varies. 4.2 Front end of ECGPro This section explains about the interfaces and the features of the software. Following are the main features provided by the ECGPro which are important to the end users. 4.3 Maintenance of patient details ECGPro keeps tracking between patient name and the records of the patient using his/her index no. Therefore it store recoded data of all 12 leads in a folder which named using index no. of the patient. 4.4 Selection of leads Figure 3.8 Circuit diagram of the summing amplifier 3.5 Designing of an A/D converter & microcontroller program PIC 16F877A microcontroller has inbuilt 10 bit A/D converter. Therefore here it has been used as the A/D converter. But this allows analogue signals between 0 5 V only. Therefore it needs an external circuit to shift the DC level of the input signals Control signals for lead selector circuit is produce by the microcontroller. This used standard parallel port of the PC to communicate with the computer. According to the control information given by the computer software, the microcontroller sends control signals to the lead selector. Software package 4.1 Control software Since the CEM is going to control via a computer, it needs to have software which is capable of controlling the device. ECGPro is the control software which customary design to fulfill the requirements. ECGPro has been design by using MATLAB 5.3 software tool. MATLAB has been choosing as the development tool, since it easy to handle when it comes to digital signal processing. This feature enables user to select the mode of recording. There are two types of recording modes. Automatic Recording Manual Recording If some one use automatic recording mode (AR mode), then the ECGPro will record all 12 leads and display them on the graphs shown right side of the interface. When it using manual recording mode (MR mode), software enables user to select the lead which is interested and display it on the graph corresponds to the selected lead. 4.5 Printing and storing of data Recorded data stored in a separate folder for each patient. It replaces old data of the patient by new data. Therefore software carries only latest data of a patient. Actually there is no need of past data since ECG Analyzer maintains past diagnostic reports. These recoded ECG patterns can be printed into a normal A4 size paper using a normal printer attached to the PC. Figure 4.1 and Figure 4.2 shows two interface of the ECGPro software.

6 Figure 4.1 This is the main interface of ECGPro. By using Help button one can get the help window which give a guide to use the interface Figure 4.2 Expanded version of the interface of ECGPro. 12 graphs shown in the right side of the figure will display the wave patterns of corresponding leads after recording.

7 Discussion & Conclusion 5.1 Issues encountered Designing of signal detector and filters Detection of low voltage signal and reduction of noise were main issues that encountered during designing phase of the signal detector. Detecting of low voltage signal was overcome by using an instrumentation amplifier. Since instrumentation amplifier consists of two noninverting buffer amplifiers, it provides high input impedance to the signal. Therefore it enables to detect the input signal without any distortion. Since signal & noise levels were comparable, noise contribution to the output was quite significant. Therefore some filtering technique required to eliminate noise from output signal to get the desired wave form. Since the frequency range of ECG wave in the range of 0.05 Hz 100 Hz (Section 1.4), higher frequency noises can eliminate using low pass filter at the output terminal of the instrumentation amplifier. Elimination of 50 Hz frequency component from AC power signal was quite a difficult task, since if an electronic circuit used to filter the 50 Hz frequency component, it can also be eliminates required frequency components (difficult to design an accurate notch filter). Therefore 50 Hz component was filtered using digital filters. Digital filters were also used to reduce the base line fluctuations due to the existence of low frequency components (frequencies les than 0.05 Hz) Port access under Windows 2000/NT Data transferring via parallel port Parallel port used for data transferring between microcontroller and the computer software. To transfer data without errors it needs to have synchronization between two ends. Initially, the synchronization was made by using delay routines. But it failed since the delay time depends on performances of the computer. Failure of delay routines gives rise to introduce a hardware handshaking technique between two ends. Introduction of hardware handshaking enables to transfer data without a distortion. 5.2 Recording of ECG Lead I was the electrode configuration used for recording. The main reason to choose Lead I for recording was the safety of the test person. Since Lead I use only the limb leads (right arm and left arm electrodes), in the case of a leakage of a small current, it is not going to affect the patient (test person). Therefore before recoding of leads like V1, V2, and V3 etc, it is required to electrically isolate patient from the circuit. Figure 4.3 and 4.6 represents two ECGs recorded using Lead I configuration. 5.3 Conclusions Designing of signal detector circuit and lead selector circuit was successfully completed; also signal detector circuit tested by recording an ECG for Lead I. Designing and development of software package was also completed successfully. Therefore it can be conclude that the aims and objectives of project have been achieved. To communicate with parallel port MATLAB needs to use external dynamic link library files. Therefore pin.dll and pout.dll (by Dr. R Lelwala, 1998) used for this purpose. But for Windows 2000/NT, this causes an access privilege violation error. Therefore system file called UserPort.SYS (Dale Roberts, 1995) used to enable user mode programs access to I/O ports Designing of software package MATLAB 5.3 used to develop the software package. Although MATLAB have been used for problem solving purposes, haven t used it for software developing. Therefore it was a challenging task at the beginning and it took a considerable time to familiar with the software tool.

8 References [1] WILLS J. TOMPKINSON, Biomedical Digital Signal Processing (2001) [2] PAUL HOROWITZ And WINFIELD HILL, The Art Of Electronics (3 rd Edition) [3] WILLI KAISER And MARTIN FINDEIS, International Journal Of BioElectromagnetism IJBEM. SECTION: NOVEL SIGNAL PROCESSING METHODS FOR EXERCISE ECG [4] AKBAR M SAYEED, PAUL LANDER AND DOUGLAS L. JONES, Improved TimeFrequency Filtering Of Signal Averaged Electrocardiograms, Journal Of Electrocardiogram, Vol. 28, Pp. 5358, Supplement 1995